Overhaul of Linux License Could Have Broad Impact

The rules governing the use of most free software programs will be revised for the first time in 15 years, in an open process that begins today.

Free software, once regarded as a tiny counterculture in computing, has become a mainstream technology in recent years, led by the rising popularity of programs like the GNU Linux operating system.

Industry analysts estimate that the value of hardware and software that use the Linux operating system is $40 billion. And Linux has become a competitive alternative to Microsoft's Windows, especially in corporate data centers.

So the overhaul of the General Public License, which covers Linux and many other free software programs, is an issue of far greater significance today than before.

"The big boys, corporations and governments, have far more reason to be interested and concerned this time," said Eben Moglen, general counsel to the Free Software Foundation, which holds the license, commonly known as the G.P.L.

The process will also be closely watched for how the new G.P.L. will take account of software patents, which have exploded among proprietary software developers since 1991, the last time the license was revised.

A document that describes the principles and timeline for updating the G.P.L., as well as the process for public comment and resolving issues, was to be posted today at www.gplv3.fsf.org. A first draft will be presented at a conference at the Massachusetts Institute of Technology, scheduled for Jan. 16 and 17.

A second draft is expected by summer. If a third draft is required, it should be completed by the fall of 2006. The process, Mr. Moglen said, could involve comments from thousands of corporations and individuals, but the Free Software Foundation will make the final judgments.

The revision process promises to be intriguing because of the man behind the G.P.L., Richard Stallman, the founder of the Free Software Foundation.

The G.P.L., according to Mr. Stallman, is an effort to use copyright law to protect what he calls the "four basic freedoms of software" - the unrestricted right to use, study, copy and modify software. The license also requires that any modifications be redistributed with the same unrestricted rights.

Mr. Stallman is renowned as both a brilliant computer programmer and a person of emphatic views on matters of software. At the Artificial Intelligence Lab at M.I.T. in the 1980's, Mr. Stallman began writing a free version of the proprietary Unix operating system, which he called GNU, and he distributed his work free.

Mr. Stallman, working as a lone craftsman, wrote a huge amount of code for the operating system and software tools for building it. But he had not gotten around to designing the "kernel" of the free operating system - the core of the program, controlling a computer's most basic functions.

In 1991, Linus Torvalds, a university student in Finland, wrote a kernel and wrapped much of the GNU code around it. The completed operating system attracted a following of programmers, working collaboratively to debug and improve the program. The operating system became known as Linux, and the networked style of work was called open source.

In Mr. Stallman's view, proprietary software is an unwarranted restriction on the freedom of information. The revision of the G.P.L., he said, is "part of something bigger - part of the long-term effort to liberate cyberspace." Software patents, he said, are "utterly insane."

For Microsoft's part, Steven A. Ballmer, its chief executive, has called the G.P.L. a "cancer."

Yet in his way, Mr. Stallman is also quite pragmatic. Proprietary software applications can run on Linux without restrictions, which is important for the survival of Linux as a viable alternative to commercial operating systems.

Mr. Stallman also acknowledged that patent rights are a sticky issue for free software, because any attempt in the G.P.L. to counteract the spread of patented software could backfire by making it difficult for free and proprietary software to run on the same computer.

"Patents are a serious issue we have to address, but we have limited leverage," he said. "Sometimes, if you push too hard, you end up pushing yourself back instead of hurting your adversary."

Kids Gone Wild

Chip Wass

Students Ace State Tests, but Earn D's From U.S.

After Tennessee tested its eighth-grade students in math this year, state officials at a jubilant news conference called the results a "cause for celebration." Eighty-seven percent of students performed at or above the proficiency level.

But when the federal government made public the findings of its own tests last month, the results were startlingly different: only 21 percent of Tennessee's eighth graders were considered proficient in math.

Such discrepancies have intensified the national debate over testing and accountability, with some educators saying that numerous states have created easy exams to avoid the sanctions that President Bush's centerpiece education law, No Child Left Behind, imposes on consistently low-scoring schools.

A comparison of state test results against the latest National Assessment of Educational Progress, a federal test mandated by the No Child Left Behind law, shows that wide discrepancies between the state and federal findings were commonplace.

In Mississippi, 89 percent of fourth graders performed at or above proficiency on state reading tests, while only 18 percent of fourth graders demonstrated proficiency on the federal test. Oklahoma, North Carolina, Alabama, Georgia, Alaska, Texas and more than a dozen other states all showed students doing far better on their own reading and math tests than on the federal one.

The chasm is significant because of the compromises behind the No Child Left Behind law. The law requires states to participate in the National Assessment - known to educators as NAEP (pronounced nape) - the most important federal measure of student proficiency.

But in a bow to states' rights, states are allowed to use their own tests in meeting the law's central mandate - that schools increase the percentage of students demonstrating proficiency each year. The law requires 100 percent of the nation's students to reach proficiency - as each state defines it - by 2014.

States set the stringency of their own tests as well as the number of questions students must answer correctly to be labeled proficient. And because states that fail to raise scores over time face serious sanctions, there is little incentive to make the exams difficult, some educators say.

"Under No Child Left Behind, the states get to set the proficiency bar wherever they like, and unfortunately most are setting it quite low," said Michael J. Petrilli, a vice president of the Thomas B. Fordham Foundation, which generally supports the federal law.

"They're telling the public in their states that huge numbers of students are proficient, but the NAEP results show that's not the case," Mr. Petrilli said.

Other educators and experts give different reasons for the discrepancy between state and federal test results. A Standard & Poor's report this fall listed many reasons for such differences, among them that the National Assessment is a no-stakes test, while low scores on state tests lead to sanctions against schools.

The report noted that the National Assessment is given to a sampling of students, whereas schools administer state tests to nearly all students. The tests serve different purposes, with the federal one giving policy makers a snapshot of student performance across the nation, while state tests provide data about individual performance. Because of these differences, some state officials say it is unfair to compare the test results.

But the report by Standard & Poor's, which has a division that analyzes educational data, also noted some states' tests are just easier.

G. Gage Kingsbury, director of research at the Northwest Evaluation Association, a nonprofit group that administers tests in 1,500 districts nationwide, said states that set their proficiency standards before No Child Left Behind became law had tended to set them high.

"The idea back then was that we needed to be competitive with nations like Hong Kong and Singapore," he said. "But our research shows that since N.C.L.B. took effect, states have set lower standards."

Not all have a low bar. In South Carolina, Missouri, Wyoming and Maine, state results tracked closely with the federal exam.

South Carolina is a state that set world-class standards, Mr. Kingsbury said. The math tests there are so difficult that only 23 percent of eighth graders scored at or above the proficiency level this year, compared with 30 percent on the federal math test. South Carolina officials now fear that such rigor is coming back to haunt them.

"We set very high standards for our tests, and unfortunately it's put us at a great disadvantage," said Inez M. Tenenbaum, the state superintendent of education. "We thought other states would be high-minded too, but we were mistaken."

South Carolina's tough exams make it harder for schools there to show the annual testing gains demanded by the federal law.

This year less than half of the state's 1,109 schools met the federal law's benchmark for the percentage of students showing proficiency, a challenge that will get tougher each year. As a result, legislators are pushing to lower the state's proficiency standard, Ms. Tenenbaum said, an idea she opposes.

Because of the discrepancies, several prominent educators are now calling for a system of national testing that counts, like those at the heart of educational systems in England, France and Japan.

"We need national standards and national tests," said Diane Ravitch, a professor at New York University who is a former member of the National Assessment's board. "I conclude that states are just looking to make everybody feel good."

Ms. Tenenbaum too says the differences among states have convinced her of the need for a national test. "I think we should all just take the NAEP," she said. "Get it out of the states' hands."

But Representative John A. Boehner, chairman of the House Committee on Education and the Work Force, defended states' rights to define proficiency as they see fit and said that over time comparisons with the federal test would force them to draw up better tests.

"The bright lights of accountability are going to shine on the states who are kidding themselves," said Mr. Boehner, Republican of Ohio.

The battle lines have long been sharp in the testing debate. Most corporate leaders favor national testing, said Susan Traiman, a director at the Business Roundtable, a group that represents corporate executives.

Opponents include liberal groups that dislike all standardized testing; the testing industry itself, which has found lucrative profits in writing new exams for all 50 states; and political conservatives who fiercely resist any intrusion on states' rights to control curricula and tests.

Margaret Spellings, the secretary of education, says that the comparison of state and federal tests provides useful information. "It allows us to shine a light," she said. "This is a truth-in-advertising type deal."

But Ms. Spellings has declined to criticize states whose tests appear to overstate the percentage of their students who are proficient. The law leaves it to states to calibrate their accountability systems, including how difficult they make their exams, she said. "We're not going to sit up in Washington and look at all those moving parts," Ms. Spellings said.

The National Assessment uses three performance levels to classify student results: advanced, which denotes superior performance; proficient, which indicates that students have "demonstrated competency" and basic, which indicates students have attained only "partial mastery of prerequisite knowledge and skills." Many students also score below basic, which the National Assessment's governing board does not classify as an achievement level.

On Oct. 19, the day the federal results were released, Ms. Spellings urged reporters to compare the percentage of students performing at the proficiency level on state tests with the percentage of students performing at the basic level on the federal test.

Many state officials said they also preferred that comparison, which greatly softens the discrepancies. In Tennessee, for instance, the 66-point gap between the federal and state results in eighth-grade math shrinks to just 26 points if the state results are compared with the federal measure of basic skills.

"NAEP's basic is comparable to our proficient," said Kim Karesh, a spokeswoman for the Tennessee Department of Education. "Now whether Tennessee's test is stringent enough is something that we're reviewing constantly. Nobody here would say we have a perfect test."

Officials in many other states whose scores differed sharply from those of the National Assessment cried foul over the very idea of comparing the results.

"The comparison to NAEP is not fair," said Mitch Edwards, a spokesman for the Department of Education in Alabama, where 83 percent of fourth-grade students scored at or above proficient on the state's reading test while only 22 percent demonstrated proficiency on the federal reading test. "Making comparisons to the NAEP becomes very difficult without giving the impression that some states are not measuring up to others or to the nation."

In Georgia, 83 percent of eighth graders scored at or above proficient on state reading tests, compared with just 24 percent on the federal test. "Kids know the federal test doesn't really count," said Dana Tofig, a spokesman for the State Department of Education. "So it's not a fair comparison; it's not apples to apples."